A100X
NVIDIA
80GB VRAM · 2040 GB/s bandwidth · 79.6 FP16 TFLOPS · 300W TDP
The A100X has 80GB of VRAM with 2040 GB/s memory bandwidth and 79.6 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (632 tok/s), Qwen 2.5 7B (332 tok/s), Llama 3.1 8B (315 tok/s). Models larger than ~136B parameters won't fit even at Q4. Electricity cost is approximately $32/month at 300W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 632 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 332 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | comfortable | 315 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | comfortable | 105 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | comfortable | 92 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | comfortable | 77 tok/s |
| Llama 3.3 70B | 70.6B | 35 GB | comfortable | 35 tok/s |
| Qwen 2.5 72B | 72.7B | 36 GB | comfortable | 34 tok/s |
| Llama 3.1 405B | 405B | 202 GB | won't fit | |
| DeepSeek R1 671B | 671B | 336 GB | won't fit |
Similar GPUs
| GPU | VRAM | BW | TFLOPS | TDP |
|---|---|---|---|---|
| A100 SXM4 80 GB | 80GB | 2040 GB/s | 78.0 | 400W |
| A800 SXM4 80 GB | 80GB | 2040 GB/s | 78.0 | 400W |
| H100 CNX | 80GB | 2040 GB/s | 215.4 | 350W |
| H100 PCIe 80 GB | 80GB | 2040 GB/s | 204.9 | 350W |
| H800 PCIe 80 GB | 80GB | 2040 GB/s | 204.9 | 350W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.